CN1806445A - Coding apparatus, coding method, and codebook - Google Patents
Coding apparatus, coding method, and codebook Download PDFInfo
- Publication number
- CN1806445A CN1806445A CNA2004800167945A CN200480016794A CN1806445A CN 1806445 A CN1806445 A CN 1806445A CN A2004800167945 A CNA2004800167945 A CN A2004800167945A CN 200480016794 A CN200480016794 A CN 200480016794A CN 1806445 A CN1806445 A CN 1806445A
- Authority
- CN
- China
- Prior art keywords
- code
- code book
- code word
- integer
- book
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/102—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the element, parameter or selection affected or controlled by the adaptive coding
- H04N19/124—Quantisation
- H04N19/126—Details of normalisation or weighting functions, e.g. normalisation matrices or variable uniform quantisers
-
- H—ELECTRICITY
- H03—ELECTRONIC CIRCUITRY
- H03M—CODING; DECODING; CODE CONVERSION IN GENERAL
- H03M7/00—Conversion of a code where information is represented by a given sequence or number of digits to a code where the same, similar or subset of information is represented by a different sequence or number of digits
- H03M7/30—Compression; Expansion; Suppression of unnecessary data, e.g. redundancy reduction
- H03M7/40—Conversion to or from variable length codes, e.g. Shannon-Fano code, Huffman code, Morse code
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/10—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding
- H04N19/169—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding
- H04N19/187—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using adaptive coding characterised by the coding unit, i.e. the structural portion or semantic portion of the video signal being the object or the subject of the adaptive coding the unit being a scalable video layer
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N19/00—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals
- H04N19/90—Methods or arrangements for coding, decoding, compressing or decompressing digital video signals using coding techniques not provided for in groups H04N19/10-H04N19/85, e.g. fractals
- H04N19/94—Vector quantisation
Abstract
A coding apparatus (500) includes: a fourth layer codebook that shows N number of (N is a natural number) codes indicating uniquely respective N integers that increment one by one; and first to third layer codebooks that show M number of (M is a natural number that satisfies M<N) codes indicating uniquely respective M integers that are a subset of the N integers, and codes a digital signal using any one of the first to fourth layer codebooks. Therefore, the coding apparatus (500) does not need to do rescaling to even when switching one of the first to fourth layer codebooks into another of them.
Description
Technical field
The present invention relates to a kind of code device, a kind of coding audio signal and the coding method of vision signal and code book that is used to encode effectively of being used for.
Background technology
With huffman coding (Huffman coding) method is the multiple effective variable length decoding method of representative, is adopted by audio signal and video signal coding.For example: be commonly called MPEG-2AAC (Moving Picture Coding Experts Group-2, AdvancedAudio Coding: moving image encoding MPEG-2, advanced audio) etc. international standard ISO/IEC 13818-7 is just as a kind of audio-frequency signal coding method and well-known.In this coding method, adopt multiple huffman coding method, this is because their code efficiency is very high.
Under the situation of AAC, in order to obtain the compression of data volume, (ModifiedDiscrete Cosine Transform: improved discrete cosine transform) coefficient is scaled and be quantified as integer, and then, the coefficient that quantized is represented as the Huffman code word for MDCT.Here, in the AAC huffman coding, the Huffman code word is assigned in a definite minimum and maximum quantification integer value scope all integers as representative points for quantization.The distribution of of this sort Huffman code word means by evenly spaced representative point, and the distance table between a certain minimum and maximum value is shown Euclidean distance.As a result, can only in each space, inerrably control quantization error, and not consider to quantize the distribution of object.In addition, use a plurality of Huffman code books, provide for the still less bit consumption of greater amount error and for the selection between more bit consumptions of small quantization error more with different quantified precisions.
Fig. 1 illustrates the structure of utilizing conventional AAC method to come the code device 100 of coding audio signal.Code device 100 comprises unit for scaling 101, quantifying unit 102, code word generation unit 103, code book preservation unit 104 and code stream generation unit 105.By the digital audio and video signals in the time domain of coming the acquisition of sampled analog audio signal with fixed frequency, be divided into the sampling of many group fixed number by the time interval with rule, described sampling is converted to the MDCT coefficient as the data in the frequency domain by MDCT, be used as supplied with digital signal Y then and be input in the code device 100.Supplied with digital signal Y is imported into unit for scaling 101, and these unit for scaling 101 usefulness are called as a certain gain A of the frequency band of scale factor band for each, and the MDCT coefficient that is contained in each frequency band is carried out convergent-divergent.Unit for scaling 101 is exported scalar gain A in code stream generation unit 105, and will output in the quantifying unit 102 as the MDCT coefficient X that scaled results obtained.Quantifying unit 102 is used predetermined transformation for mula, the MDCT coefficient X that quantization zooming is crossed.More particularly, under the situation of AAC method, the MDCT coefficient of representing with floating number rounded off obtains integer value, is exactly its quantification.Code word generation unit 103 is encoded to the MDCT coefficient that quantized, and is exported its code word C in code stream generation unit 105 with reference to the code book that is stored in the code book preservation unit 104.Code stream generation unit 105 adds header etc. among gain A and the code word C in this scale factor band to, and they are exported as code stream, and wherein, gain A is from unit for scaling 101 inputs, and code word C is from 103 inputs of code word generation unit.Preserve in the unit 104 at code book, stored a plurality of code books with different quantified precisions.If carry out coding with reference to the code book that has low quantified precision in a plurality of code books that are stored in the code book preservation unit 104, the code stream bit number can be reduced, but as cost, introduces bigger quantization error.On the contrary, if encode with reference to the code book with higher quantization resolution, quantization error can be reduced, but as cost, needs more code stream bit number.In a word, the code book of this code device 100 by selecting to be suitable for encoding from a plurality of code books with different quantified precisions improved code efficiency.
Fig. 2 is illustrated in the example of the code book of storing in the code book preservation unit 104 of conventional code device 100 as shown in Figure 1 with different quantified precisions 201 to 204.In code book 201, quantized value can be arbitrary integer " 1 ", " 0 " and " 1 " between maximum " 1 " and minimum value " 1 ", and the code word of quantized value is expressed as binary number with 0 and 1.In code book 202, quantized value can be arbitrary integer " 2 ", " 1 ", " 0 ", " 1 " and " 2 " between maximum " 2 " and minimum value " 2 ", and the code word of quantized value is expressed as binary number with 0 and 1.In code book 203, quantized value can be arbitrary integer " 3 ", " 2 ", " 1 ", " 0 ", " 1 ", " 2 " and " 3 " between maximum " 3 " and minimum value " 3 ", and the code word of quantized value is expressed as binary number with 0 and 1.In code book 204, quantized value can be arbitrary integer " 4 ", " 3 ", " 2 ", " 1 ", " 0 ", " 1 ", " 2 ", " 3 " and " 4 " between maximum " 4 " and minimum value " 4 ", and the code word of quantized value is expressed as binary number with 0 and 1.These code words only are some examples, and the present invention is not limited to this.
Fig. 3 illustrates each code book that is used for reference to figure 2, and the distribution of the representative point (quantized value X ') that the object X that will quantize is quantized is promptly as the set of integers of representative point.As shown in Figures 2 and 3, code book 201 illustrates code word and is assigned on Three Represents point " 1 ", " 0 " and " 1 ".In this case, as the object X that will quantize
1The MDCT coefficient scaled, thereby become the value between maximum " 1 " and minimum value " 1 " (but being not limited to those values) greatly.As mentioned above, under the situation that reference code book 201 is encoded, be expression representative point separately, code word only needs three values, therefore only needs the seldom bit of number.Code book 202 illustrates code word and is assigned to as quantized value X
2' five representative points " 2 ", " 1 ", " 0 ", " 1 " and " 2 " on.In this case, as the object X that will quantize
2The MDCT coefficient scaled, thereby become the value between maximum " 2 " and minimum value " 2 " (but being not limited to those values) greatly.As mentioned above, under the situation that reference code book 202 is encoded, for representing five quantized value X
2', need have than three quantized value X of expression in code book 201
1' situation under the code word of multi-bit more.Similarly, under the situation that reference code book 203 is encoded, code word is assigned to seven quantized value X of from " 3 " to " 3 "
3' on, and as the object X that will quantize
3The MDCT coefficient scaled, thereby become the value between maximum " 3 " and minimum value " 3 " greatly.Under the situation that reference code book 204 is encoded, code word is assigned to nine quantized value X of from " 4 " to " 4 "
4' on, and as the object X that will quantize
4The MDCT coefficient scaled, thereby become the value between maximum " 4 " and minimum value " 4 " greatly.Under the situation of AAC as audio coding method, using as mentioned above, method quantizes and encodes.
In addition, as shown in Figure 2, design codebooks 201 to 204 makes the code word of shorter bit form be assigned on the quantized value of frequent appearance.
Yet in conventional method, the object that will quantize only is quantized into a plurality of representative points, and these representative points evenly are distributed in 0 being on the positive direction and negative direction at center, and do not consider the actual distribution situation of object.For example, obvious by Fig. 3, corresponding to the representative point 301 to 304 of code book 201 to 204, being distributed in 0 with identical distance is on the positive direction and negative direction at center.The object that will quantize distributes with identical distance, and this is common on statistics, but hardly they be that the center is even spaced apart and never be partial to positive direction or negative direction with 0 always.
In addition, in conventional method, the object that uses a plurality of original quantifications of Huffman code with different quantified precisions to quantize is to improve code efficiency.Yet when when a code book switches to another code book with different quantified precisions, the object that will quantize needs by convergent-divergent again, to be fit to the precision of another code book.Fig. 4 is illustrated in conventional code device 100 to use under the situation that code book with different quantified precisions encodes the flow chart of the example of encoding process process.Here, as an example of coding method, will the situation of at first encoding with reference to the code book with maximum quantization precision be described.Suppose in this case, when the bit number in the scale factor band surpasses when having the result's that the code book of maximum quantization precision encodes reference value as use, encode with reference to code book with low one-level quantified precision, then, such coding is repeated to carry out, and the bit number in this scale factor band becomes reference value or littler or do not exist till the code book with lower quantified precision.At first, the individual MDCT coefficient (y of " N " in each scale factor band (wherein, N is a natural number)
1, y
2, y
3..., y
N) be imported into (S401) in the unit for scaling 101.The MDCT coefficient of 101 pairs of inputs of unit for scaling carries out convergent-divergent, makes its value drop into the range of representative points interior (S402) in the code book with maximum quantization precision.For example, under the situation that reference code book 204 is encoded, as equation-4＜(y
k/ A)=x
k＜+4 (k=1 wherein, 2 ..., when n) all these MDCT coefficients being set up, unit for scaling 101 with gain A to " k " individual MDCT coefficient y
kCarry out convergent-divergent.The MDCT coefficient x that 102 pairs of convergent-divergents of quantifying unit are crossed
kRound off, and code word generation unit 103 is read code word C from code book 204
k, code word C wherein
kThe quantized value x that obtains by rounding off will be assigned to
k'.Code word generation unit 103 calculates N code word (C in this scale factor band
1, C
2, C
3..., C
N) total bit number (S403), and judge that whether the total bit number calculate is target bit or less than target bit (S404).
Total bit number target bit number of the code word in this scale factor band in step S404 or more hour ("Yes" among the S404), code word generation unit 103 is the code word (C of the code book ID of this scale factor band that is used for encoding (perhaps sequence number etc.) and this scale factor band
1, C
2, C
3..., C
N) output in the code stream generation unit 105.Code stream generation unit 105 output code flows comprise: from the code book ID of code word generation unit 103 outputs and the code word (C this scale factor band
1, C
2, C
3..., C
N); Scalar gain A (S405) from unit for scaling 101 outputs.MDCT coefficient in next scale factor band is imported in the code device 100, and code device 100 begins above-mentioned processing from step S401.
N in this scale factor band in step S404 code word (C
1, C
2, C
3..., C
N) total bit number during greater than target bit ("No" among the S404), code word generation unit 103 order unit for scaling 101 are in same scale factor band, to MDCT coefficient (y
1, y
2, y
3..., y
N) carry out convergent-divergent again, make its value drop in the value scope of representative point of code book with low one-level quantified precision.According to this instruction, unit for scaling 101 returns processing and the zoom factor (S402) again among the step S402.More particularly, has the purpose of encoding for reference, as equation-3＜(y than the code book 203 of code book 204 low one-level quantified precisions
k/ A)=x
k＜+3 (k=1 wherein, 2 ..., when n) all these MDCT coefficients being set up, unit for scaling 101 usefulness gain A are to " k " individual MDCT coefficient y
kCarry out convergent-divergent.The MDCT coefficient x that 102 pairs of convergent-divergents of quantifying unit are crossed
kRound off, code word generation unit 103 is read code word C from code book 203
k, this code word C wherein
kThe quantized value x that obtains by rounding off will be assigned to
k'.N the code word (C that code word generation unit 103 calculates in this scale factor band
1, C
2, C
3..., C
N) total bit number (S403), and judge whether the total bit number calculate is target bit or littler (S404).N in this scale factor band code word (C
1, C
2, C
3..., C
N) total bit number under current still situation greater than target bit, code device 100 returns the processing among the step S402, revise the value of gain A, and the processing of repeating step S402 to S404 is up to not existing code book with lower quantified precision or the total bit number in scale factor band to become target bit or littler.
As mentioned above, in order to repeat convergent-divergent, not only need N the MDCT coefficient that is contained in each scale factor band cut apart, and each real number value of this N MDCT coefficient and the value scope of representative point need be compared.These are cut apart and relatively need a large amount of calculating, and this has brought problem to coding.
Opposite with background mentioned above, the purpose of this invention is to provide a kind of code device and a kind of decoding device, wherein this code device allows more effective coding, and does not increase the processing load that is caused by convergent-divergent again.
Summary of the invention
In order to address this problem, code device according to the present invention is a kind of code device that is used for the original coded digital signal of reference code, comprising: preserve the unit, be used to preserve first code book and at least one second code book; Coding unit, be used for reference to being kept at one of a plurality of code books of preserving the unit, replace digital signal with code word, wherein, first code book is illustrated in the one-to-one relationship between N (N is a natural number) individual code word and N the continuous integral number, described at least one second code book is illustrated in the one-to-one relationship between M (M is the natural number less than N) individual code word and M the integer, and coding unit uses by the integer value of digital signal indication to come with reference to being kept at one of first and second code books of preserving in the unit.
According to aforesaid the present invention, used dissimilar code books, focal attention is in this fact, and the representative point that promptly is used to quantize always need not represented by even representative point spaced apart in the value scope between maximum and minimum value.More particularly, in conventional code book, representative point is represented by all integer values in the value scope between maximum and minimum value, and in code book of the present invention, they can be represented by fewer purpose integer value, be less than all integer values in the value scope between maximum and minimum value.As a result, the distribution of the representative point in same scope can be used as a new characteristic and adds in the code book of the present invention.In addition, the variation of numbers of representative points means a plurality of code books with different quantified precisions, and wherein said numbers of representative points is less than the number of all integers in the value scope between maximum and minimum value.Therefore, switch quantified precision, can improve code efficiency by bit number according to the code stream that will generate.In addition, because concerning all code books, maximum is identical with minimum value, even switch to another code book with another quantified precision from a code book with a quantified precision, also need not to carry out convergent-divergent again, like this, just can reduce and switch the required amount of calculation of code book.As a result, can expect that code device of the present invention provides various advantages, when being implemented in LSI and the software, can reduce power consumption as it.
As the more information about the present techniques background, the Japanese patent application No.2003-170287 that on June 16th, 2003 submitted to discloses, and comprises specification, accompanying drawing and claim, all is incorporated herein by reference at this.
The accompanying drawing summary
From the description of the present invention being carried out below in conjunction with accompanying drawing, these and other objects of the present invention, advantage and characteristic will become apparent, wherein description of drawings one particular embodiment of the present invention.In these accompanying drawings:
Fig. 1 illustrates and utilizes conventional ACC method to come the structure of the code device of coding audio signal;
Fig. 2 is illustrated in the example of a plurality of code books of storing in the code book preservation unit of conventional code device as shown in Figure 1 with different quantified precisions;
Fig. 3 illustrates the distribution that is used for the representative points for quantization (quantized value X ') that each code book with reference to figure 2 quantizes the object X that will quantize;
Fig. 4 is illustrated in the flow chart that conventional code device uses an example of the encoding process process under the situation that code book with different quantified precisions encodes;
Fig. 5 is the block diagram that the structure of the code device among first embodiment is shown;
Fig. 6 illustrates a kind of form and the relation between the quantified precision of a group code in this of the distribution deflection "+2 " of representative point wherein of the distribution of the representative points for quantization behind the convergent-divergent;
Fig. 7 illustrate the representative points for quantization behind the convergent-divergent distribution another kind of form and wherein representative point evenly be distributed in relation between the quantified precision in " 0 " code book on every side;
Fig. 8 illustrates the another kind of form and the relation between the quantified precision of a group code in this of the distribution deflection " 2 " of representative point wherein of the distribution of the representative points for quantization behind the convergent-divergent;
Fig. 9 illustrates the example of code book of the present embodiment of representative quantization method as shown in Figure 6;
Figure 10 is the flow chart that the operation of the code device in the present embodiment is shown;
Figure 11 is the block diagram that the structure of the decoding device among first embodiment is shown;
Figure 12 illustrates the example of Two dimensional Distribution of the representative point of the code book that is used for the coding that the second embodiment code device carries out; And
Figure 13 is illustrated in the example such as the Two dimensional Distribution of the representative point of the code book that uses in the such conventional coding method of ACC.
Embodiment
Encoding apparatus and decoding apparatus in the embodiments of the invention are described below with reference to the accompanying drawings.
(first embodiment)
Fig. 5 is the block diagram that the structure of the code device 500 among first embodiment is shown.In Fig. 5, identical reference number be assigned in the code device 500 with as shown in Figure 1 code device 100 in identical part.Code device 500 is code devices that one of a plurality of code books with different quantified precisions of a kind of reference are encoded.Here, these code books have different distributions of representative points for quantization in identical integer value scope, and because therefore the numbers of representative points difference in that integer range has different quantified precisions.Code device 500 comprises quantifying unit 102, code stream generation unit 105, unit for scaling 501, code word generation unit 503 and code book preservation unit 504.Except when code word generation unit 503 unit for scaling 501 when a code book switches to another code book does not carry out again outside the convergent-divergent, unit for scaling 501 is identical with unit for scaling 101.Identical with the situation of unit for scaling 101, unit for scaling 501 uses handle such as the such digital signal that will encode of (a plurality of) MDCT coefficient Y such as equatioies 1 to resolve into (a plurality of) factor X of (a plurality of) object that gain A and conduct will quantize.
Y=AX...... equation 1
Here, A and X can be integer values, also can be floating-point numerical value.For simplicity, X is a floating-point numerical value.Quantifying unit 102 is quantized into integer value X ' to the X that is imported by rounding off etc.Code word generation unit 503 is searched for the code word that corresponds respectively to the integer value X ' that is imported in code book 504, and it is exported as code word C.Yet, code word generation unit 503 is with the difference of code word generation unit 103, the former is by obtaining from average and the deviation (dispersion) of the quantized value X ' of quantifying unit 102 inputs, check the distribution bias of the object that will quantize, and from many group codes bases, select one group according to distribution bias, in order to the coded quantization value.Every group code corresponding to the distribution bias of the object that will quantize originally comprises a plurality of code books with different quantified precisions.Code word generation unit 503 at first uses the code book that has the maximum quantization precision in selected code book group to encode, total bit number of the code word of calculating in a scale factor band, and judge whether the total bit number that is calculated is target bit or littler.When it is a target bit or more hour, code word generation unit 503 is with ID or be used for discerning the codebook number of the code book that is used to encode and the code word C that obtains by coding outputs to code stream generation unit 105.When total bit number of the code word in this scale factor band during greater than target bit, code word generation unit 503 switches to the code book that another has low level encoder quantified precision with this code book.At this moment, different with code word generation unit 103, code word generation unit 503 does not order unit for scaling 501 to carry out convergent-divergent again.By this way, code word generation unit 503 repeated encodings also calculate total bit number, until the total bit number of the code word in this scale factor band become target bit or littler or do not have code book with lower quantified precision till.105 utilizations of code stream generation unit generate bit stream from the gain A (perhaps indicating the index of gain A) of unit for scaling 501 outputs, from the code word C of code word generation unit 503 outputs and the ID that is used to discern code book.Code book preserve unit 504 preserve have different quantified precisions many group codes this, and every group code originally has the distribution bias of the oneself object that will quantize, and preserves unit 104 unlike code book, only preserve have different quantified precisions a group code originally.
Because the characteristics of present embodiment are how to construct many group codes originally, therefore will the code book that use in the code device 500 of constructing as mentioned above be elaborated by comparing with conventional code book.To use Fig. 6, Fig. 7, Fig. 8 and Fig. 9 that quantization method and coding method in the present embodiment are described.To use Fig. 2 and Fig. 3 that routine techniques is described.In addition, situation in the present embodiment is that value (object that will the quantize) X that is input in the quantifying unit 502 is the one dimension value.
Fig. 6 illustrates a kind of form 601 to 604 and the relation between the quantified precision of a group code in this of the distribution deflection "+2 " of representative point wherein of the distribution of the representative points for quantization behind the convergent-divergent.Fig. 7 illustrate the representative points for quantization behind the convergent-divergent distribution another kind of form 701 to 704 and wherein representative point evenly be distributed in relation between the quantified precision in " 0 " code book on every side.Fig. 8 illustrates the another kind of form 801 to 804 and the relation between the quantified precision of a group code in this of the distribution deflection " 2 " of representative point wherein of the distribution of the representative points for quantization behind the convergent-divergent.Each shaded block among Fig. 6, Fig. 7 and Fig. 8 is indicated each representative point.
Fig. 6 is illustrated in the distribution of representative point and is partial in the code book group of "+2 ", the distribution situation 601 to 604 of representative points for quantization, and code word is assigned on these representative points for quantization.In the distribution 601 of representative point, code word is assigned on the integer value " 4 ", " 2 " and " 0 " of the representative point nine values of from " 4 " to " 4 ".In distribution 602, code word is assigned on representative point " 4 ", " 2 ", " 1 ", " 0 " and " 4 ".In distribution 603, code word is assigned on representative point " 4 ", " 2 ", " 1 ", " 0 ", " 1 ", " 3 " and " 4 ".Distributing in 604, code word is assigned on all nine integer values from " 4 " to " 4 ", and this situation with conventional code book 204 is identical.
Fig. 7 is illustrated in representative point and evenly is distributed in " 0 " code book group on every side, the distribution 701 to 704 of representative points for quantization, and code word is assigned on these representative points for quantization.In the distribution 701 of representative point, code word is assigned on the integer value " 2 ", " 0 " and " 2 " of representative point.In distribution 702, code word is assigned on representative point " 4 ", " 2 ", " 0 ", " 2 " and " 4 ".In distribution 703, code word is assigned on representative point " 4 ", " 2 ", " 1 ", " 0 ", " 1 ", " 2 " and " 4 ".Distributing in 704, code word is assigned on all nine integer values from " 4 " to " 4 ", and this situation with conventional code book 204 is identical.
Fig. 8 is illustrated in the distribution of representative point and is partial in the code book group of " 2 ", the distribution 801 to 804 of representative points for quantization, and code word is assigned on these representative points for quantization.In the distribution 801 of representative point, code word is assigned on the integer value " 0 ", " 2 " and " 4 " of representative point.In distribution 802, code word is assigned on representative point " 4 ", " 0 ", " 1 ", " 2 " and " 4 ".In distribution 803, code word is assigned on representative point " 4 ", " 3 ", " 1 ", " 0 ", " 1 ", " 2 " and " 4 ".Distributing in 804, code word is assigned on all nine integer values from " 4 " to " 4 ", and this situation with conventional code book 204 is identical.
In Fig. 6, Fig. 7 and Fig. 8 the distribution 601,701 of representative point and 801 each in, be three as the number of the shaded block of representative point, this is identical with the number of shaded block in the distribution 301 of Fig. 3.In Fig. 6, Fig. 7 and Fig. 8 the distribution 602,702 of representative point and 802 each in, be five as the number of the shaded block of representative point, this is identical with the number of shaded block in the distribution 302 of Fig. 3.In Fig. 6, Fig. 7 and Fig. 8 in the distribution 603,703 and 803 of representative point, be seven as the number of the shaded block of representative point, this is identical with the number of shaded block in the distribution 303 of Fig. 3.In Fig. 6, Fig. 7 and Fig. 8 the distribution situation 604,704 of representative point and 804 each in, in the number of representative point and its distribution 304 with Fig. 3 of distributing is identical.On the number of the shaded block that is primarily focused on here, this number is indicated the number of code word in the code book of each quantified precision.
In Fig. 6, Fig. 7, Fig. 8 and Fig. 3, the code word of similar number in the code book is described, these code books have following separately corresponding to the distribution form of quantified precision: the distribution 601,701,801 and 301 of (1) representative point; (2) distribute 602,702,802 and 302; (3) distribute 603,703,803 and 303; (4) distribute 604,704,804 and 304.In other words, this means in Fig. 6, Fig. 7, Fig. 8 and Fig. 3 that the code book that is used to quantize with identical quantified precision has the code word of similar number.
Then, the representative point branch that only focuses among Fig. 6 plants, the shaded block in 601 of the distributing shaded block in 602 that always distributes, the shaded block in 602 of the distributing shaded block in 603 that always distributes, the shaded block in 603 of the distributing shaded block in 604 that always distributes.In brief, in distribution 604, all pieces in maximum and minimum value scope all are shaded blocks.This sets up equally for the distribution among Fig. 7 and Fig. 8 701 to 704 and 801 to 804.The shaded block in 701 and 801 of the distributing shaded block in 702 and 802 that always distributes, the shaded block in 702 and 802 of the distributing shaded block in 703 and 803 that always distributes, the shaded block in 703 and 803 of the distributing shaded block in 704 and 804 that always distributes.In brief, in distribution 704 and 804, all pieces in maximum and minimum value scope all are shaded blocks.Here, using wherein, quantized value is called as " ground floor quantification " according to the quantification that the code book that distributes shown in 601,701 or 801 as distributing carries out.Similarly, the quantification of use distribution 602,702 or 802 is called as " second layer quantification ", uses the quantification of distribution 603,703 or 803 to be called as " the 3rd layer of quantification ", uses the quantification of distribution 604,704 or 804 to be called as " the 4th layer of quantification ".In addition, the code book that is used for this first, second, third and the 4th layer of quantification is called as " ground floor code book ", " second layer code book ", " the 3rd layer of code book " and " the 4th layer of code book " respectively.
Situation below any quantization methods of four layers of ground floors to the can be applied to, thereby the scaled value that becomes between maximum " 4 " and minimum value " 4 " of the object that promptly will quantize.Therefore, even quantization method changes over from any method of four layers of ground floors to the,, also need not to carry out convergent-divergent again even just from four layers of code book of ground floor to the switches to wherein another.In addition, along with the increase of from first to the 4th number of plies, quantified precision can be amplified safely.In the 4th layer of quantization method of present embodiment, use with conventional distribution 304 in quantization method in identical code book encode.Therefore, for example under predetermined situation of encoding, need not to carry out convergent-divergent again, this difference and conventional method owing on highest resolution, encoding by quantization method with full accuracy.All will be done just switches code book.In addition, shift to an earlier date under the given situation in the signal distributions of the object that will quantize, the quantized value of each layer is not limited to the integer value sequence, compares with routine techniques, can also be set more neatly according to distribution.
Fig. 9 illustrates an example of the code book in the present embodiment, representative quantization method as shown in Figure 6.Fig. 2 illustrates conventional code book, representative quantization method as shown in Figure 3.
In code book 901, as the quantized value of representative points for quantization can be any three integers (as " 4 ", " 2 " and " 0 ") between maximum " 4 " and minimum value " 4 ", and the code word of all these three quantized values is expressed as binary value (as 111,100 and 0) with 0 and 1.In code book 902, as the quantized value of representative points for quantization can be any five integers (as " 4 ", " 2 ", " 1 ", " 0 " and " 4 ") between maximum " 4 " and minimum value " 4 ", and the code word of all these five quantized values is expressed as binary value (as 1010,100,00,01 and 1011) with 0 and 1.In code book 903, quantized value can be any seven integers (as " 4 ", " 2 ", " 1 ", " 0 ", " 1 ", " 3 " and " 4 ") between maximum " 4 " and minimum value " 4 ", and the code word of all these seven quantized values is expressed as binary value (as 10101,10100,100,00,01,10110 and 10111) with 0 and 1.In code book 904, quantized value can be the arbitrary integer (" 4 ", " 3 ", " 2 ", " 1 ", " 0 ", " 1 ", " 2 ", " 3 " and " 4 ") between maximum " 4 " and minimum value " 4 ", and the code word of all quantized values is expressed as binary value (as 1010001,10101,1010000,100,00,01,10110,10111 and 1010010) with 0 and 1.These code words only are that the present invention is not limited to this for example.
Quantized value in the code book 904 always comprises the quantized value in the code book 903.Quantized value in the code book 903 always comprises the quantized value in the code book 902.Quantized value in the code book 902 always comprises the quantized value in the code book 901.Therefore, if use code book 902 rather than code book 901 to quantize, can the lower quantization error, perhaps avoid the increase of quantization error at least.Similarly, if use code book 903 rather than code book 902 to quantize, can the lower quantization error, perhaps avoid the increase of quantization error at least.Similarly, if use code book 904 rather than code book 903 to quantize, can the lower quantization error, perhaps avoid the increase of quantization error at least.Therefore, in the present embodiment, when switching code book,, just need not to carry out convergent-divergent again as long as code device 500 convergent-divergent MDCT coefficients make its value drop in the scope between maximum " 4 " and the minimum value " 4 ".In addition, along with number of plies increase of 904 from code book 901 to code book, quantization error can be reduced.
On the other hand, it is less that Fig. 9 is illustrated in the bit number of the code word in the code book 901 with minimum quantified precision, and the code word bits number in having the code book of higher quantization resolution is bigger.For example, quantized value " 4 " is represented by 3 bits 111 in the code book 901 with minimum quantified precision, but is represented by 4 bits 1010 in the code book 902 with higher quantization resolution.In having the code book 903 of high quantization precision more by 10101 expressions of 5 bits, in code book 904 with maximum quantization precision by 1010001 expressions of 7 bits.As mentioned above, along with codebook number increase of 904 from code book 901 to code book, the number of representative point increases, and therefore, bit consumption increases usually.In other words, can quantize as coding by using code book, reduce bit consumption, perhaps can quantize as coding by using code book with more number representative points for quantization with less number representative points for quantization, come the lower quantization error, bit consumption can increase although it is so.
Figure 10 is the flow chart that the operation of the code device 500 in the present embodiment is shown.At first, the individual signal value (y of " N " in each frequency band (N is a natural number)
1, y
2, y
3..., y
N) be imported into (S1001) in the unit for scaling 501,501 couples of signal value (y of this unit for scaling
1, y
2, y
3..., y
N) carry out convergent-divergent, make its value drop in the predetermined value scope (S1002).More particularly, 501 couples of signal value (y of unit for scaling
1, y
2, y
3..., y
N) carry out convergent-divergent, make the signal value x that convergent-divergent is crossed
k=y
k/ A (wherein, k=1,2 ..., N) drop in the predetermined scope, that is, for example in the scope between maximum " 4 " and minimum value " 4 ".Quantifying unit 102 is with predetermined method, for example, will count 1 and cast out remaining part greater than 1/2 part, to the signal value X behind the convergent-divergent
kRound off.Code word generation unit 503 is analyzed the quantized value X that obtains by rounding off
k' distribution (S1003).More particularly, it calculates quantized value x in this frequency band
k' (k=1,2 ..., average N) and distribution.This average illustrates quantized value x
k' deviation where concentrate on.This distribution illustrates quantized value X in this frequency band
k' depart from the degree at deviation center.Code word generation unit 503 switches the code book group to the most suitable quantized value X according to analysis result
k' distribution a group code this (S1004).Code word generation unit 503 is preserved the unit 504 from code book and is read the code book of all layers that belong to suitable code book group, and reads code word C from every layer code book
k, code word C
kTo be assigned to quantized value X
k' (k=1,2 ..., N) on.Code word generation unit 503 is also with the code word C that reads
kBe assigned to N quantized value X in this frequency band
k' (k=1,2 ..., N) on, and calculate total bit number (S1005) of every layer of code word in this frequency band.
Here, suppose from the processing of four layers of ground floors to the is parallel and carry out.More particularly, in code word generation unit 503, (1) is used to carry out the code word C that processing unit that ground floor handles is read ground floor
K1, code word C
K1To be assigned to quantized value X
k' (k=1,2 ..., N).(2) with handle (1) parallel, be used to carry out the code word C that processing unit that the second layer handles is read the second layer
K2, code word C
K2To be assigned to quantized value X
k' (k=1,2 ..., N) on.(3) with handle (2) parallel, the processing unit that is used to carry out the 3rd layer of processing is read the 3rd layer code word C
K3, code word C
K3To be assigned to quantized value X
k' (k=1,2 ..., N) on.(4) with handle (3) parallel, the processing unit that is used to carry out the 4th layer of processing is read the 4th layer code word C
K4, code word C
K4To be assigned to quantized value X
k' (k=1,2 ..., N) on.Then, code word generation unit 503 calculates " N " individual code word (C in this scale factor band
1, C
2, C
3..., C
N) total bit number.More particularly, (1) is used to carry out the processing unit that ground floor handles and calculates ground floor code word C in this frequency band
K1(k=1,2 ..., total bit number N).(2) with handle (1) parallel, be used to carry out the processing unit that the second layer handles and calculate second layer code word C in this frequency band
K2(k=1,2 ..., total bit number N).(3) with handle (2) parallel, the processing unit that is used to carry out the 3rd layer of processing calculates the 3rd layer of code word C in this frequency band
K3(k=1,2 ..., total bit number N).(4) with handle (3) parallel, the processing unit that is used to carry out the 4th layer of processing calculates the 4th layer of code word C in this frequency band
K4(k=1,2 ..., total bit number N).Code word generation unit 503 is selected certain one deck code book, and wherein the total bit number that calculates for this layer is a target bit or littler.Perhaps, be under target bit or the littler situation at the total bit number that does not have arbitrary layer, code word generation unit 503 is selected ground floor code books (S1006).
The code book that code word generation unit 503 utilization is selected in step S1006 (S1007) is encoded to the signal value in the frequency band, and code book ID of the signal value in the frequency band that is used to encode (perhaps sequence number etc.) and the interior code word (C of frequency band
1, C
2, C
3..., C
N) output in the code stream generation unit 105.Code stream generation unit 105 output code flows comprise: from the code book ID of code word generation unit 503 outputs and the code word (C the scale factor band
1, C
2, C
3..., C
N); And from the scalar gain A (S1008) of unit for scaling 501 output.After this, the signal value in the next frequency band is transfused to code device 500, begins aforesaid identical processing procedure from step S1001.
As mentioned above, according to the code device in the present embodiment 500, when each code word generation unit 503 switches code book, need not to carry out convergent-divergent again, so the processing load in the code device 500 can be lowered.In addition, because code device 500 has been preserved the many groups of code books with different distributions of representative points for quantization, therefore the distribution at quantized value shifts to an earlier date under the given situation, can rely on its distribution to select optimum code book, like this, can realize accurate more the quantification and coding.In addition, because every group code originally has a plurality of code books with different quantified precisions, therefore can carry out efficient coding more.
Figure 11 is the block diagram that the structure of the decoding device 1100 among first embodiment is shown.Decoding device 1100 in first embodiment is a kind of decoding devices, is used for the code stream of decoding and being encoded by the code device 500 of first embodiment.This decoding device 1100 comprises code stream separative element 1101, decoding unit 1102, reverse unit for scaling 1103 and code book preservation unit 1104.In decoding device 1100, bit stream is imported into the code stream separative element 1101 from the outside via recording medium or transmission path.Code stream separative element 1101 is according to predetermined bit stream syntax, isolates gain A, code word C and be used to discern the sequence number or the ID of code book from code stream.Isolated code word C and the being used for sequence number of discerning code book be imported into decoding unit 1102.Isolated gain A is imported in the reverse unit for scaling 1103.It is identical to be stored in the code book that code book preserves in the code book preservation unit 504 in code book and the code device 500 that is stored in as shown in Figure 5 in the unit 1104.Decoding unit 1102 identifies a code book that is stored in the code book preservation unit 1104 based on the sequence number that is used to discern code book.With reference to the code book that is identified, decoding unit 1102 further is decoded as quantized value X ' with code word C based on quantized value X ' that describes in this code book and corresponding code word C.Be imported in the reverse unit for scaling 1103 from the quantized value X ' of decoding unit 1102 outputs and the gain A of from code stream, separating.For example, oppositely unit for scaling 1103 amplifies the quantized value X ' of output from decoding unit 1102 with gain A, thereby obtains the value Y ' of MDCT coefficient according to equation 2 etc.
Y '=AX ' ... equation 2
As mentioned above, according to present embodiment, preserve code book in the unit 504 and change into and have any code book that the different representative points shown in Fig. 6 to 8 distribute as long as be stored in the code book of code device 500, the code efficiency in the code device 500 just can be enhanced, and can obviously not increase its amount of calculation.In addition, when bit stream that decoding generates in code device 500, decoding device 1100 can be based on the gain A of describing in this code stream, code word C and is used to discern the sequence number or the ID of code book, without difficulty code stream decoding is become digital signal.
(second embodiment)
In first embodiment, illustrated will quantize to as if the situation of one-dimensional signal value.Yet the object that will quantize need not one dimension always, also can be multidimensional.For for simplicity, as an example, with below with reference to Figure 12 and Figure 13 explanation will quantize to as if the situation of two-dimensional signal values.Figure 12 illustrates the example of Two dimensional Distribution of the representative point of the code book that is used for the coding that the second embodiment code device carries out.Figure 13 is illustrated in the example such as the Two dimensional Distribution of the representative point of the code book that uses in the such conventional coding method of ACC.It is x that Figure 12 and Figure 13 schematically show at trunnion axis, vertical axis be y and the bivector that comprises component x and y (x, y) under the given situation, the distribution of representative points for quantization.Here, for example, the quantized value of twice continuous sampling can be applied to this two-dimensional vector components.
In the present embodiment, for example, use quantization method as shown in figure 12.More particularly, in in ground floor distribution 1201, the second layer distribute 1202 and the 3rd layers of distribution 1203 any, the maximum of x and y component all is identical with minimum value,-3≤x≤3 and-3≤y≤3, and the number that has only representative point distributes 1203 increase and increases along with the number of plies distributes 1201 to the 3rd layers from ground floor.Under this structure, in case the object that will quantize is scaled and find that they are contained in any zone of the 3rd layer of distribution 1203, just can select to have the code book of the layer of high coding efficiency, thereby use the value of scaled like this mistake to encode, and need not it is carried out convergent-divergent again.For example, object that will quantize and representative point distribute in 1202 or always distribute in 1201 coupling mutually of ground floor at the second layer.Under the sort of situation, can judge in second layer distribution 1202 or ground floor distribution 1201, each object that will quantize is near which shadow region (representative point), and the representative point in definite nearest shadow region is as the quantized value that will encode.Situation about distributing as one dimension along with the increase of the sequence number of the layer that is used to quantize, also increases from distribute 1201 to the 3rd layers the distribute number of 1203 representative point of ground floor.Therefore, in order to reduce quantization error,, can use the code book of representative point, but bring the increase of bit consumption thereupon with greater number for quantification.On the other hand,,, can use the code book that has than the representative point of peanut, but bring may increasing of quantization error thereupon for quantification in order to reduce bit consumption.In addition, the representative points for quantization in the code book can be created arbitrarily based on the distribution of the object that will quantize, and is not subjected to evenly restriction spaced apart, therefore can expect the raising of code efficiency.
On the contrary, according to the conventional coding method as AAC, the distribution of representative points for quantization, for example, ground floor distributes 1301, by by determining maximum and the minimum value on each in x direction of principal axis and y direction of principal axis, and the distance between maximum and the minimum value is divided into three zones, and is divided into nine zones.Scaled and when being divided in these nine zones any when the object that will quantize, it is used that regional representative point (quantized value) and quantizes.For example, distribute in 1301 at ground floor, when the maximum behind the convergent-divergent and minimum value are " 1 " and " 1 " respectively, 9 points (1 ,-1), (1,-0), (1,1), (0 ,-1), (0,0), (0,1), (1 ,-1), (1,0) and (1,1) are used as representative points for quantization.Similarly, in second layer distribution 1302, the distance between maximum and the minimum value is divided into five zones, and in the 3rd layer of distribution 1303, the distance between maximum and the minimum value is divided into seven zones.In conventional code book, under any situation, the multidimensional of representative points for quantization distributes similar with the situation of its one dimension distribution, is represented by even representative point spaced apart.In addition, under the situation that switches to another code book from a code book, (for example, switch to second layer code book), must carry out re-quantization, and therefore increased amount of calculation from the ground floor code book.
It should be noted that in second embodiment the situation of the representative point distribution bias of a kind of form as shown in figure 12 only has been described, still, the present invention is not limited to this situation.Much less, identical with the situation of first embodiment, for the distribution of various ways, the present invention can preserve many group codes originally.
In addition, illustrated in the above-described embodiments that the input signal that will encode (object that will quantize) is the situation of audio signal, still, the present invention is not limited to this situation, as long as coding method of the present invention can be applied to these signals, can import the signal of any kind of so.For example, can incoming video signal.What import will encoded signals be under the situation of vision signal, DCT coefficient for input does not carry out the convergent-divergent processing, but, can use a plurality of code books that are contained in the code device to realize coding more efficiently according to the different distributions of the object that will quantize.
Though above only exemplary embodiments more of the present invention are described in detail, but will understanding at an easy rate, those skilled in the art can carry out many modifications in this exemplary embodiment, and not from deviating from novel teachings of the present invention and advantage in essence.Therefore, all such modifications are attempted within the scope of the invention involved.
Industrial applicibility
Zuo be contained in personal computer with compression and coded audio and video signal functions, The code device of the Zhong such as the broadcasting station of personal digital assistant (PDA), digital broadcasting, mobile phone, Code device according to the present invention is of great value.
Zuo is the code that is contained in the code device Zhong of Yong Yu compression and coded audio and vision signal This and Zuo are the decoding device Zhong that is contained in the bit stream of Yong Yu decoding You code device coding Code book, code book according to the present invention is of great value.
Claims (23)
1, a kind of code device is used for comprising with reference to the code book coded digital signal:
Preserve the unit, be used to preserve first code book and at least one second code book; And
Coding unit is used for replacing digital signal with reference to being kept at one of described code book of described preservation unit with code word,
Wherein, described first code book is illustrated in the one-to-one relationship between N (N is a natural number) individual code word and N the continuous integral number,
Described at least one second code book is illustrated in the one-to-one relationship between M (M is the natural number less than N) individual code word and M the integer, and
Described coding unit uses by the integer value of described digital signal indication to come with reference to being kept at one of described first and second code books in the described preservation unit.
2, code device as claimed in claim 1,
Wherein, a described M integer is a subclass of a described N integer.
3, code device as claimed in claim 2,
Wherein, preserve under the situation of a plurality of second code books in described preservation unit, described second code book is illustrated in M1 respectively, M2 ... (M1, M2, ... be natural number) individual code word and M1, M2 ... the one-to-one relationship between the individual integer, described M1, M2 ... be numeral different from each other, and described M1, M2 ... individual integer comprises a set of integers public to all described second code books.
4, code device as claimed in claim 3,
Wherein, at least one the 3rd code book is also preserved in described preservation unit,
Described at least one the 3rd code book is illustrated in the one-to-one relationship between K (K is the natural number less than N) individual code word and K the integer, and a described K integer is a subclass of a described N integer,
Described at least one the 3rd code book comprises a set of integers in the described K integer, and described set of integers is different from the set of integers that is included in any one described second code book, and
Preserve under the situation of a plurality of the 3rd code books in described preservation unit, described trigram originally is illustrated in K1, K2 respectively, ... (K1, K2 ... be natural number) individual code word and K1, K2, ... the one-to-one relationship between the individual integer, described K1, K2, ... be numeral different from each other, and described K1, K2 ... individual integer comprises a set of integers public to all described the 3rd code books.
5, code device as claimed in claim 4 also comprises:
Unit for scaling is used for the signal value that convergent-divergent is imported, and makes described signal value drop into and comprises in the predetermined value scope of a described N integer;
Quantifying unit, the signal value that is used for respectively described convergent-divergent being crossed are quantized to the digital signal value in the value scope of a described N integer; And
Selected cell is used for selecting described at least one second code book or described at least one the 3rd code book according to the signal value collection of described quantification,
Wherein, described coding unit is replaced the signal value of described quantification with reference to a code book in described at least one that select second code book or at least one the 3rd code book with code word.
6, code device as claimed in claim 5,
Wherein, described selected cell calculates the average and the distribution of described digital signal value, checks the signal value collection of described quantification, and based on the result of described inspection, selects described at least one second code book or described at least one the 3rd code book.
7, code device as claimed in claim 6,
Wherein, described coding unit comprises:
The bit count unit is used for calculating the bit number of code word of described replacement of the digital signal of each sampling that comprises predetermined number; And
Judging unit is used to judge whether the described bit number that calculates is target bit or littler, and
With reference to described first code book and the bit number of the described code word of replacing is under described target bit or the littler situation, described coding unit is with reference to described first code book; Under the situation of bit number greater than described target bit of the described code word of replacing with reference to described first code book, described coding unit is with reference to a code book in described at least one that select second code book or at least one the 3rd code book.
8, code device as claimed in claim 7,
Wherein, described coding unit is with reference to a code book in described at least one that select second code book or at least one the 3rd code book, the bit number of the described code word in a described code book is no more than under the situation of described target bit, a described code book comprises M1, M2, ... a middle maximum number code word or K1, K2 ... a middle maximum number code word.
9, code device as claimed in claim 3,
Wherein, described coding unit comprises:
The bit count unit is used for calculating the bit number of code word of described replacement of the digital signal of each sampling that comprises predetermined number; And
Judging unit is used to judge whether the described bit number that calculates is target bit or littler, and
With reference to described first code book and the bit number of the described code word of replacing is under described target bit or the littler situation, described coding unit is with reference to described first code book; Under the situation of total bit number greater than described target bit of the described code word of replacing with reference to described first code book, described coding unit is with reference to a code book in described at least one second code book.
10, code device as claimed in claim 9,
Wherein, described coding unit is with reference to a code book in described at least one second code book, and the bit number of the described code word in a described code book is no more than under the situation of described target bit, and a described code book comprises M1, M2 ... a middle maximum number code word.
11, a kind of coding method, be used for coming coded digital signal with reference to first code book that is kept at the preservation unit and a code book at least one second code book, described coding method comprises coding step, described coding step is used for reference to being kept at one of described code book of described preservation unit, replace digital signal with code word
Wherein, described first code book illustrates the one-to-one relationship between N (N is a natural number) individual code word and N the continuous integral number,
Described at least one second code book illustrates the one-to-one relationship between M (M is the natural number less than N) individual code word and M the integer; And
In described coding step, by using by the integer value of described digital signal indication to come with reference to being kept at one of described first and second code books in the described preservation unit.
12, coding method as claimed in claim 11,
Wherein, a described M integer is a subclass of a described N integer.
13, coding method as claimed in claim 12,
Wherein, preserve under the situation of a plurality of second code books in described preservation unit, described second code book is illustrated in M1 respectively, M2 ... (M1, M2, ... be natural number) individual code word and M1, M2 ... the one-to-one relationship between the individual integer, described M1, M2 ... be numeral different from each other, and described M1, M2 ... individual integer comprises a set of integers public to all described second code books.
14, coding method as claimed in claim 13,
Wherein, at least one the 3rd code book is also preserved in described preservation unit,
Described at least one the 3rd code book is illustrated in the one-to-one relationship between K (K is the natural number less than N) individual code word and K the integer, and a described K integer is a subclass of a described N integer,
Described at least one the 3rd code book comprises a set of integers in the described K integer, and described set of integers is different from the set of integers that is included in any one described second code book, and
Preserve under the situation of a plurality of the 3rd code books in described preservation unit, described trigram originally is illustrated in K1, K2 respectively, ... (K1, K2 ... be natural number) individual code word and K1, K2, ... the one-to-one relationship between the individual integer, described K1, K2, ... be numeral different from each other, and described K1, K2 ... individual integer comprises a set of integers public to all described the 3rd code books.
15, coding method as claimed in claim 14 also comprises:
The convergent-divergent step is used for the signal value that convergent-divergent is imported, and makes described signal value drop into and comprises in the predetermined value scope of a described N integer;
Quantization step, the signal value that is used for respectively described convergent-divergent being crossed are quantized to the digital signal value in the value scope of a described N integer; And
Select step, be used for selecting described at least one second code book or described at least one the 3rd code book according to the signal value collection of described quantification,
Wherein, in described coding step,, replace the signal value of described quantification with code word with reference to a code book in described at least one that select second code book or at least one the 3rd code book.
16, coding method as claimed in claim 15,
Wherein, in described selection step, calculate the average and the distribution of described digital signal value, check the signal value collection of described quantification, and, select described at least one second code book or described at least one the 3rd code book based on the result of described inspection.
17, coding method as claimed in claim 16,
Wherein, described coding step comprises:
The bit count step is used for calculating the bit number of code word of described replacement of the digital signal of each sampling that comprises predetermined number; And
Determining step is used to judge whether the described bit number that calculates is target bit or littler, and
In described coding step, with reference to described first code book and the bit number of the described code word of replacing is under described target bit or the littler situation, with reference to described first code book; Under the situation of bit number greater than described target bit of the described code word of replacing, with reference to a code book in described at least one that select second code book or at least one the 3rd code book with reference to described first code book.
18, coding method as claimed in claim 17,
Wherein, in described coding step, with reference to a code book in described at least one that select second code book or at least one the 3rd code book, the bit number of the described code word in a described code book is no more than under the situation of described target bit, a described code book comprises M1, M2 ... a middle maximum number code word or K1, K2 ... a middle maximum number code word.
19, coding method as claimed in claim 13,
Wherein, described coding step comprises:
The bit count step is used for calculating the bit number of code word of described replacement of the digital signal of each sampling that comprises predetermined number; And
Determining step is used to judge whether the described bit number that calculates is target bit or littler, and
In described coding step, with reference to described first code book and the bit number of the described code word of replacing is under described target bit or the littler situation, with reference to described first code book; Under the situation of total bit number greater than described target bit of the described code word of replacing, with reference to a code book in described at least one second code book with reference to described first code book.
20, coding method as claimed in claim 19,
Wherein, in described coding step, with reference to a code book in described at least one second code book, the bit number of the described code word in a described code book is no more than under the situation of described target bit, a described code book comprises M1, M2 ... a middle maximum number code word.
21, a kind of decoding device is used for will becoming digital signal by the code stream decoding that code word is represented with reference to code book, comprising:
Extraction unit is used for extracting with the lower part from the sampling of each predetermined number of input code flow: the described code word of representing described digital signal; Gain is used to amplify the described digital signal that obtains as the result of the described code word of decoding; And identifying information, be used for being identified in and convert described code word to code book that the integer number signal value uses;
Preserve the unit, be used to preserve a plurality of code books;
Decoding unit is used for discerning a code book in a plurality of code books that are kept at described preservation unit, and with reference to the described code book that is identified, converts the described code word that extracts to described integer number signal value;
Amplifying unit is used to utilize the described gain that extracts, and amplifies the described digital signal value that obtains as the result of described conversion.
22, a kind of coding/decoding method is used for will becoming digital signal by the code stream decoding that code word is represented with reference to a code book that is kept in a plurality of code books of preserving the unit, comprising:
Extraction step is used for extracting with the lower part from the sampling of each predetermined number of input code flow: the described code word of representing described digital signal; Gain is used to amplify the described digital signal that obtains as the result of the described code word of decoding; And identifying information, be used for being identified in and convert described code word to code book that the integer number signal value uses;
Decoding step is used for discerning a code book in a plurality of code books that are kept at described preservation unit, and with reference to the described code book that is identified, converts the described code word that extracts to described integer number signal value;
Amplification procedure is used to utilize the described gain that extracts, and amplifies the described digital signal value that obtains as the result of described conversion.
23, code book, it is by with reference to being used for replacing digital signal with code word, and described code book is illustrated in the one-to-one relationship between M (M is the natural number less than N) individual code word in the individual continuous integral number scope of N (N is a natural number) and M the integer,
Wherein, a described M integer is a subclass of a described N integer.
Applications Claiming Priority (2)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2003170287 | 2003-06-16 | ||
| JP170287/2003 | 2003-06-16 |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1806445A true CN1806445A (en) | 2006-07-19 |
| CN100493199C CN100493199C (en) | 2009-05-27 |
Family
ID=33549419
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2004800167945A Expired - Fee Related CN100493199C (en) | 2003-06-16 | 2004-06-07 | Coding apparatus, coding method, and codebook |
Country Status (4)
| Country | Link |
|---|---|
| US (1) | US7657429B2 (en) |
| JP (1) | JP4382090B2 (en) |
| CN (1) | CN100493199C (en) |
| WO (1) | WO2004112400A1 (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015000416A1 (en) * | 2013-07-04 | 2015-01-08 | 华为技术有限公司 | Frequency domain envelope vector quantization method and apparatus |
| WO2020164153A1 (en) * | 2019-02-15 | 2020-08-20 | Oppo广东移动通信有限公司 | Method for determining configuration parameter, and terminal device and network device |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20080012740A1 (en) * | 2006-07-06 | 2008-01-17 | Board Of Trustees Of Michigan State University | Source-aware non-uniform information transmission with minimum distortion |
| EP1883067A1 (en) * | 2006-07-24 | 2008-01-30 | Deutsche Thomson-Brandt Gmbh | Method and apparatus for lossless encoding of a source signal, using a lossy encoded data stream and a lossless extension data stream |
| US9031851B2 (en) * | 2007-09-20 | 2015-05-12 | Lg Electronics Inc. | Method and an apparatus for processing a signal |
| JP5846220B2 (en) * | 2012-01-17 | 2016-01-20 | ソニー株式会社 | Encoding device, encoding method, decoding device, decoding method, and program |
| US9240853B2 (en) | 2012-11-16 | 2016-01-19 | Huawei Technologies Co., Ltd. | Systems and methods for sparse code multiple access |
| US9712817B1 (en) * | 2013-02-15 | 2017-07-18 | Apple Inc. | Lossless video coding systems and methods |
| US10531432B2 (en) | 2015-03-25 | 2020-01-07 | Huawei Technologies Co., Ltd. | System and method for resource allocation for sparse code multiple access transmissions |
| US9419770B2 (en) | 2014-03-31 | 2016-08-16 | Huawei Technologies Co., Ltd. | Method and apparatus for asynchronous OFDMA/SC-FDMA |
| US10701685B2 (en) | 2014-03-31 | 2020-06-30 | Huawei Technologies Co., Ltd. | Method and apparatus for asynchronous OFDMA/SC-FDMA |
| WO2017071586A1 (en) | 2015-10-30 | 2017-05-04 | Huawei Technologies Co., Ltd. | System and method for high-rate sparse code multiple access in downlink |
| US20220156982A1 (en) * | 2020-11-19 | 2022-05-19 | Nvidia Corporation | Calculating data compression parameters |
Family Cites Families (24)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5341457A (en) * | 1988-12-30 | 1994-08-23 | At&T Bell Laboratories | Perceptual coding of audio signals |
| JP2874363B2 (en) | 1991-01-30 | 1999-03-24 | 日本電気株式会社 | Adaptive encoding / decoding method |
| US5408234A (en) * | 1993-04-30 | 1995-04-18 | Apple Computer, Inc. | Multi-codebook coding process |
| JP2951861B2 (en) | 1994-12-28 | 1999-09-20 | シャープ株式会社 | Image encoding device and image decoding device |
| GB9509831D0 (en) * | 1995-05-15 | 1995-07-05 | Gerzon Michael A | Lossless coding method for waveform data |
| EP0756386B1 (en) * | 1995-07-27 | 2003-09-17 | Victor Company Of Japan, Limited | Method and apparatus for coding a digital, acoustic signal |
| US5956674A (en) * | 1995-12-01 | 1999-09-21 | Digital Theater Systems, Inc. | Multi-channel predictive subband audio coder using psychoacoustic adaptive bit allocation in frequency, time and over the multiple channels |
| US5818877A (en) * | 1996-03-14 | 1998-10-06 | The Regents Of The University Of California | Method for reducing storage requirements for grouped data values |
| SG74707A1 (en) | 1998-03-06 | 2000-08-22 | Gen Electric | Cyclic bis-phosphites and compositions |
| DE19840835C2 (en) * | 1998-09-07 | 2003-01-09 | Fraunhofer Ges Forschung | Apparatus and method for entropy coding information words and apparatus and method for decoding entropy coded information words |
| JP3352406B2 (en) * | 1998-09-17 | 2002-12-03 | 松下電器産業株式会社 | Audio signal encoding and decoding method and apparatus |
| US6157328A (en) * | 1998-10-22 | 2000-12-05 | Sony Corporation | Method and apparatus for designing a codebook for error resilient data transmission |
| JP3323175B2 (en) * | 1999-04-20 | 2002-09-09 | 松下電器産業株式会社 | Encoding device |
| US6446037B1 (en) * | 1999-08-09 | 2002-09-03 | Dolby Laboratories Licensing Corporation | Scalable coding method for high quality audio |
| US6678653B1 (en) * | 1999-09-07 | 2004-01-13 | Matsushita Electric Industrial Co., Ltd. | Apparatus and method for coding audio data at high speed using precision information |
| KR100856398B1 (en) * | 2000-05-17 | 2008-09-04 | 삼성전자주식회사 | Variable-length coding and decoding method and apparatus using plural mapping tables |
| US6944226B1 (en) | 2000-10-03 | 2005-09-13 | Matsushita Electric Corporation Of America | System and associated method for transcoding discrete cosine transform coded signals |
| JP2002223443A (en) | 2001-01-24 | 2002-08-09 | Yamaha Corp | Transcoding method and transcoder |
| JP3639216B2 (en) * | 2001-02-27 | 2005-04-20 | 三菱電機株式会社 | Acoustic signal encoding device |
| JP2002311997A (en) | 2001-04-16 | 2002-10-25 | Matsushita Electric Ind Co Ltd | Audio signal encoder |
| JP3636094B2 (en) * | 2001-05-07 | 2005-04-06 | ソニー株式会社 | Signal encoding apparatus and method, and signal decoding apparatus and method |
| JP3778087B2 (en) * | 2002-01-18 | 2006-05-24 | 富士ゼロックス株式会社 | Data encoding apparatus and data decoding apparatus |
| US7724827B2 (en) * | 2003-09-07 | 2010-05-25 | Microsoft Corporation | Multi-layer run level encoding and decoding |
| US7426462B2 (en) * | 2003-09-29 | 2008-09-16 | Sony Corporation | Fast codebook selection method in audio encoding |
-
2004
- 2004-06-07 JP JP2006516835A patent/JP4382090B2/en not_active Expired - Fee Related
- 2004-06-07 WO PCT/JP2004/008258 patent/WO2004112400A1/en active Application Filing
- 2004-06-07 CN CNB2004800167945A patent/CN100493199C/en not_active Expired - Fee Related
- 2004-06-07 US US10/557,149 patent/US7657429B2/en not_active Expired - Fee Related
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2015000416A1 (en) * | 2013-07-04 | 2015-01-08 | 华为技术有限公司 | Frequency domain envelope vector quantization method and apparatus |
| US9805732B2 (en) | 2013-07-04 | 2017-10-31 | Huawei Technologies Co., Ltd. | Frequency envelope vector quantization method and apparatus |
| RU2635069C2 (en) * | 2013-07-04 | 2017-11-08 | Хуавэй Текнолоджиз Ко., Лтд. | Device and method of quantizing vectors of envelope frequencies |
| US10032460B2 (en) | 2013-07-04 | 2018-07-24 | Huawei Technologies Co., Ltd. | Frequency envelope vector quantization method and apparatus |
| WO2020164153A1 (en) * | 2019-02-15 | 2020-08-20 | Oppo广东移动通信有限公司 | Method for determining configuration parameter, and terminal device and network device |
| CN111837344A (en) * | 2019-02-15 | 2020-10-27 | Oppo广东移动通信有限公司 | Method for determining configuration parameters, terminal equipment and network equipment |
| CN111837344B (en) * | 2019-02-15 | 2022-01-11 | Oppo广东移动通信有限公司 | Method for determining configuration parameters, terminal equipment and network equipment |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100493199C (en) | 2009-05-27 |
| JP4382090B2 (en) | 2009-12-09 |
| US20060245489A1 (en) | 2006-11-02 |
| WO2004112400A1 (en) | 2004-12-23 |
| JP2006527961A (en) | 2006-12-07 |
| US7657429B2 (en) | 2010-02-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| EP2282310B1 (en) | Entropy coding by adapting coding between level and run-length/level modes | |
| EP1913698B1 (en) | Efficient coding and decoding of transform blocks | |
| CN1126268C (en) | Method of and device for coding digital information signal | |
| CN103067022B (en) | A kind of integer data lossless compression method, decompression method and device | |
| CN100493199C (en) | Coding apparatus, coding method, and codebook | |
| US8558724B2 (en) | Coding method, coding appartaus, decoding method, decoding apparatus, program, and recording medium | |
| CN102017634A (en) | Multi-level representation of reordered transform coefficients | |
| KR100959837B1 (en) | Adaptive method and system for mapping parameter values to codeword indexes | |
| JP2004258603A (en) | Entropy encoding adapting encoding between level mode and run length/level mode | |
| Goyal et al. | On optimal permutation codes | |
| US6255968B1 (en) | Data compression method and apparatus | |
| CN1615590A (en) | Data compression and expansion of a digital information signal | |
| KR20110033154A (en) | Method for counting vectors in regular point networks | |
| CN1158051A (en) | Method and apparatus for approximating contour image of object in video signal | |
| WO2019198383A1 (en) | Encoding device, decoding device, encoding method, decoding method, program, and recording medium | |
| CN1918563A (en) | Method and apparatus for transforming a digital audio signal and for inversely transforming a transformed digital audio signal | |
| Kieffer et al. | Progressive lossless image coding via self-referential partitions | |
| CN110771045A (en) | Encoding device, decoding device, encoding method, decoding method, and program | |
| Mikhael et al. | Energy-based split vector quantizer employing signal representation in multiple transform domains | |
| Perlmutter et al. | Vector quantization with zerotree significance map for wavelet image coding | |
| JP4971965B2 (en) | Encoding device, decoding device, encoding method, decoding method, encoding program, decoding program, and recording medium | |
| Kossentini et al. | High order entropy-constrained residual VQ for lossless compression of images | |
| JPH0472870A (en) | Picture encoding system | |
| CN1761159A (en) | Encoding and decoding device for quantized step in any module, and method | |
| CN1758757A (en) | Method and device for getting decoded data value of binary decoding |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20090527 Termination date: 20190607 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |